EXPERIMENTAL STUDIES OF THE THEORIES FOR THE MAINTENACE OF SEXUAL REPRODUCTION USING SACCHAROMYCES CEREVISIAE MEGHNA B.

Transcription

1 EXPERIMENTAL STUDIES OF THE THEORIES FOR THE MAINTENACE OF SEXUAL REPRODUCTION USING SACCHAROMYCES CEREVISIAE BY MEGHNA B. OSTASIEWSKI A Dissertation Submitted to the Graduate Faculty of WAKE FOREST UNIVERSITY GRADUATE SCHOOL OF ARTS AND SCIENCES in Partial Fulfillment of the Requirements for the Degree of DOCTOR OF PHILOSOPHY Biology December 2013 Winston-Salem, North Carolina Approved By: Clifford Zeyl, Ph.D., Advisor Kenneth Berenhaut, Ph.D., Chair David Anderson, PhD. Robert Browne, Ph.D. Miles Silman, Ph.D.

2 TABLE OF CONTENTS List of Illustrations and Tables iii List of Abbreviations...v Abstract vi Chapter 1: Introduction Chapter 2: Spatial Structure and the Evolution of Sex. 10 Chapter 3: Mating Type Switching, Segregation, and the Advantages of Sexual Reproduction in Harsh Environments Chapter 4: The Tangled Bank Chapter 5: Conclusion References..83 Curriculum Vitae...88 ii

3 LIST OF ILLUSTRATIONS AND TABLES CHAPTER 2 Table I: Forward and reverse primer sequences for replacing the IME1 gene...20 Figure 1: Replica-plating scheme 21 Figure 2: Frequencies of sexuals in the ending populations evolved on agar plates with and without rotation...24 Figure 3: Fitness of random isolates from evolved populations..25 Figure 4: Numbers of asci out of 100 cells in each population after one experimental cycle...26 Table II: Two factor ANOVA of the effects of dispersal, sporulation, and their interactions on the final frequencies of sexuals in the ending populations Table III: Three factor ANOVA of the effects of dispersal, sporulation, sexual ability, their interactions on the final fitness of isolates from populations evolved on agar plates..28 CHAPTER 3 Figure 1: Frequency of homothallics over time in populations evolved in dextrose minimal medium at 30 C...44 Figure 2: Change in frequency of homothallics over time in populations evolved in dextrose minimal, high osmolarity medium at 37 C.45 Table I: Initial and ending frequencies of homothallics in each of the four treatments..46 Figure 3: Average fitness of randomly selected isolates from final evolved populations Table II: Logistic regression on frequency of homothallics in the long-term evolution experiment. 48 Table III: ANOVA of the effect of selective environment, assay environment, and thallism on the ranked fitnesses of isolates from ending populations...49 iii

4 Table IV: ANOVA on ranked diploid fitness from week long competitions between diploid and haploid isogenic versions of the same strain..50 Figure 4: Average fitness of a diploid (MZ) strain after a week-long competition against a haploid from the same background.51 CHAPTER 4 Table I: Origins of the strains used in various experiments 63 Table II: Characteristics of the seven environments used in various experiments..64 Figure 1: Diagrams of the three types of competitions.. 65 Figure 2: Average fitness from two replicates of individual parental MZ30 and MZ32 and nine offspring from Group A in each of the seven environments Figure 3: Average fitness from two replicates of individual parental MZ30 and MZ34 and nine offspring from Group A in each of the seven environments Figure 4: Average fitness from two replicates of individual parental MZ32 and MZ34 and nine offspring from Group A in each of the seven environments...69 Table III: Two factor ANOVA on log-transformed fitness data from Group A...70 Figure 5: Fitness of individuals and strain combinations from Group A in competitions pooled across environments...71 Figure 6: Fitness of pools of YPS630 with YPS670 (Group B), ten F 1 from their cross, and ten F 2 each derived from an F 1 pooled across the seven environments..72 Figure 7: Average fitness of strains from Group B in each of the 7 environments. Top panel includes YPS630, YPS681, parent mix, and offspring mix. Bottom panel includes YPS630, YPS 670, parent mix, and offspring mix..73 iv

5 Dextrose Minimal Medium (DM) Yeast Peptone Dextrose (YPD) LIST OF ABBREVIATIONS v

6 ABSTRACT The existence and pervasiveness of sexual reproduction despite its many disadvantages, such as breaking up beneficial allele combinations and the cost of producing males, is an enduring question in evolutionary biology. It has yet to be determined which one, or which combination, of the hypothesized advantages contributes the most to its maintenance. I used lab and wild strains of the yeast Saccharomyces cerevisiae to test hypotheses regarding the importance of spatial structure, ploidy, and diversity among offspring in explaining the persistence of sexual reproduction. I found that the ability to reproduce sexually was favorable in a dispersed population, while it was not advantageous in a structured population, indicating that a lack of spatial structure can favor the maintenance of sex. Haploids were favored in a harsh environment, while diploids fared better in a benign environment, so sexual reproduction may be important in organisms such as yeast because it allows for changes in ploidy. Recombinant offspring pools had higher fitness than a pool of their parents in some but not all cases, providing qualified support for the Tangled Bank hypothesis. Taken together, these three studies suggest that evolutionary effects of sex vary with ecological setting and population structure, rather than applying more broadly to sexual taxa in general. vi

7 CHAPTER 1: INTRODUCTION Sexual reproduction, whether exclusive or sporadically, is an essential part of the life cycle of many eukaryotes; however, its evolution and maintenance have perplexed evolutionary biologists for years. Sexual females are only able to produce half as many daughters as asexual females, which results in a two-fold cost of sex. This suggests that more purely asexual species should exist because they produce twice as many offspring, a benefit that can equivalently be seen as a 100% transmission of alleles into the next generation as opposed to the 50% reduction in the number of alleles transmitted after meiosis in a sexual species. In isogamous systems (in which gametes are morphologically indistinguishable from each other and contribute equally to zygotes), by contrast, both sexes (mating types), contribute equally to zygotes, and are equally capable of reproduction. These populations are therefore exempt from the two-fold cost of sex; although other costs may apply. For example, recombination associated with sexual reproduction also has the potential to break up favorable gene combinations that would be transmitted intact during asexual reproduction. Despite these problems, sexual reproduction has persisted, and at least twenty theoretical models (Kondrashov, 1994) have been proposed to explain why. The advantages of recombination have long been considered to be vital to the explanation of the evolution and maintenance of sex. In order for recombination to offer an evolutionary advantage, populations must tend to accumulate negative linkage disequilibrium due to selection, drift, epistasis or some interaction among those factors. Linkage disequilibrium may be positive, an excess of genotypes/chromosomes with two adaptive mutations linked, or negative, an excess of genotypes with one adaptive and one 1

8 less fit allele. Positive linkage disequilibrium does not persist because selection quickly fixes those genotypes/chromosomes with both high-fitness alleles, but negative linkage disequilibrium that can accumulate in asexual populations is reduced by sexual reproduction, allowing loci to be independent of each other. The two most popular types of explanations for the persistence of sex can be categorized as environmental theories and mutational theories. Fisher (1930) hypothesized that sex facilitates the spread of advantageous alleles because it allows these alleles to escape from their initial genetic backgrounds. Recombination also allows sexual populations to incorporate more than one new favorable adaptation into subsequent generations. In contrast, asexual populations are thought to be subjected to selective sweeps (also known as periodic selection), a process by which selection for a mutation purges diversity at all loci from a population, and competitive exclusion (Hardin, 1960), which allow only transient variation and a diminished ability to respond to a changing environment. In asexual populations, adaptive mutations arising in different lines cannot be combined as they could be in sexual populations. They will instead compete with each other until only the fittest remains, and combinations of beneficial mutations can only be obtained if they occur in succession within the same lines of descent (Muller, 1932). Sexual populations do not have to wait for these successive sweeps of individual beneficial mutations as the mutations can be combined through recombination. Two main criticisms of this Fisher-Muller hypothesis are its basis in group selection, and the requirement for perpetual selection for new allele combinations. The prevalent environmental explanation, the Red Queen Hypothesis, proposes that host- 2

9 parasite interactions satisfy the need for perpetual selection for novel genotypes. When new and rare, novel host genotypes are successful because parasites have not yet counterevolved to fit the new host allele combinations. However, the initial success of those rare host genotypes reduces their subsequent fitness as their frequencies increase in the population and they become more easily targeted by parasites. This is an example of negative frequency-dependent selection, and suggests that novel allele combinations produced through sex provide an advantage in biotic interactions. Bell and Maynard Smith (1987) determined that an allele causing free recombination among loci under direct selection would provide a large short-term selective advantage in one of two mutually antagonistic species. These mutually antagonistic species are usually thought of as hosts and parasites, with parasites usually having much shorter generation times than their hosts, allowing a faster rate of evolution. Therefore, offspring whose allele combinations are less frequent in the population than those of their parents are fitter than clones because they are less vulnerable to the prevailing parasite genotypes of their generation (Hamilton, 1980). In general, environmental theories state that sex can create new gene combinations and therefore accelerate adaptation to a perpetually changing biotic or abiotic environment. Mutational hypotheses are based on the ubiquity of harmful mutations. While not all organisms are involved in perpetual arms races with parasites, all do experience harmful mutations (Kondrashov, 1988). Muller (1932) noted that at mutation-selection equilibrium the genotype with the fewest mutations in each population is rare. Although these genotypes have a fitness advantage, they exist at a low initial frequency, and are therefore few in number in small populations, so they are periodically lost due to drift. 3

10 Only sexual populations will be able to recreate those genotypes with fewer mutations. This process, known as Muller s Ratchet, undoubtedly occurs, but has two significant limitations as a theory of sex: it only applies to small populations and provides a longterm group selection, rather than individual, advantage for sexual reproduction. Kondrashov (1982) addressed both of these issues when he considered synergistic epistasis and developed the Mutational Deterministic Hypothesis, which provides an individual advantage for sex, and is more effective in larger populations that are less subject to genetic drift. Under these conditions, mutations are synergistic, with each additional deleterious mutation causes a greater decrease in fitness than the previous mutation, and in the extreme case there is a lethal threshold number of mutations. Since sexual reproduction increases the variance in mutation number among individuals, and offspring which vary more in numbers of mutations are on average fitter, this provides an advantage for sexual reproduction. This hypothesis, however, requires a deleterious mutation rate greater than 1.0 per genome per generation, and strong synergistic epistasis between mutations (which leads to the abrupt fitness drop-off above some threshold). There is empirical evidence that many eukaryotes meet the first requirement, but there is neither empirical evidence nor biological rationale for deleterious mutations being consistently synergistic. Otto and Barton (2001) simulated models with various numbers of loci under selection and modifier alleles that could increase recombination rates, so that recombination rates were variable instead of assuming that organisms were simply either sexual or asexual. They found that in smaller populations, the modifier alleles would increase in frequency (leading to an increase in recombination rate) due to negative 4

11 linkage disequilibrium caused by genetic drift. In large populations, the modifier alleles conferring recombination could theoretically increase in frequency if weak negative epistasis (which does not depend on drift) results in fewer individuals with two or more beneficial mutations than expected from their independent fitness benefits. In this case, recombination would increase the frequency of individuals with two or more beneficial alleles, and there would be fewer individuals with only one beneficial allele each, providing an advantage for the modifier allele conferring recombination. However, recombination may not be necessary to provide an advantage to sex when segregation is considered instead. In a diploid sexual population, segregation can carry a new single-copy mutation to the homozygous state, whereas in an asexual population, two separate mutations are required at the same locus (Kirkpatrick and Jenkins, 1989). Otto (2003) has even shown that in model populations with a low level of inbreeding, selection is stronger on a modifier allele that promotes segregation than on a modifier allele that promotes recombination. However, this prediction has apparently not been directly tested. Other explanations include those of generalized group selection in which asexual lineages are subject to a higher rate of extinction, or to a lower rate of speciation than sexual lineages, which allows sex to be maintained despite its short-term disadvantage (Nunney, 1989). Sex may also be necessary in some taxa, particularly mammals, due to genomic imprinting. At some loci, either the paternal or the maternal allele is silenced; therefore crucial genes expressed from paternally derived chromosomes will not be expressed in parthenogenetically derived offspring (Hurst and Peck, 1996). 5

12 Authors such as Barton and Charlesworth (1998) have compared and contrasted the many theories on sexual reproduction. Hurst and Peck (1996) reviewed the major theories and provided evidence for rapid evolution in immune system genes (support for the Red Queen model) and estimates of the mutation rate from new experiments. Others have tried to combine the mutational and environmental theories into a pluralist approach that could solve the problems of each individual theory (West et al., 1999), while still others argue that the region of parameter space in which sex is favored is likely to be large and continuous, so a general explanation of sex should be possible (Nurenberger and Gabriel, 1999). In another general summary paper, Meirmans et al. (2012) suggest that the costs and benefits of sex may be more species-specific than the general answer most often sought after. In addition to the many theoretical studies on the evolution of sex, evolution experiments are an ideal way to determine the validity of the numerous hypotheses. The yeast Saccharomyces cerevisiae is a good model organism for evolution of sex experiments because it has very short generation times, reproduces asexually in both haploid and diploid states, and can be genetically engineered using homologous recombination for direct replacement of genes by genetic markers (Sherman, 2002). Yeast generally reproduce asexually, but diploids can be induced to go through sporulation and meiosis when on starvation medium. Once back in permissive medium, the spores will germinate and then mate with cells of the opposite mating type. Yeast strains can be either heterothallic (unable to switch mating types) or homothallic (able to switch mating types through the excision and exchange of mating type cassettes from silenced copies into a site where the resident allele is expressed). Within this system, 6

13 heterothallic cells have to mate with cells from a lineage of the other mating type, but homothallic cells can switch mating type and then mate with their own daughter cells. Otherwise identical sexual and asexual populations can be constructed to determine if and when there is a direct advantage for sexual reproduction and recombination. The following paragraphs summarize these types of yeast experiments from the past two decades. Birdsell and Wills (1996) ran competitions between otherwise isogenic sexual (heterozygous at the mating type locus) and asexual (homozygous at the mating type locus) strains and found that sexual strains always out-competed the asexuals. Furthermore, with a heterozygous background the sexual strains had a competitive advantage from the start of the experiment whether or not sporulation (meiosis) was induced, but these findings are complicated by the fact that the mating type (MAT) locus has a range of pleiotropic effects and the observed advantage of MAT heterozygosity even when there had been no meiosis. In a direct competition experiment, asexuals were created by deleting the IME1 gene, which prevents them from sporulating and subsequently undergoing meiosis (Greig et al., 1998). In this study, sexual and asexual versions of three homozygote and three heterozygote strains were derived from three unrelated ancestor strains. After one round of sporulation (sex), the sexuals and asexuals of each of the six derived strains competed in a novel 37 C environment, a stressful temperature for yeast, with the prediction that the advantages of recombination, and therefore an increase in the frequency of sexuals, would only be seen in the heterozygote strains. In each of the strains, sexual reproduction had an initial cost (most likely due to the breaking up of beneficial allele 7

14 combinations, as opposed to the two-fold cost of sex, since yeast are not anisogamous), but in two of the three unrelated heterozygote strains the sexuals outcompeted the asexuals. In the remaining strains it was a matter of chance whether the sexuals or asexuals remained at the end of the experiment. Average relative fitness did increase in each of the strains. Although not in yeast, additional support for sex- increasing adaptation to a novel environment has been documented in a comparison of rates of adaptation in separate sexual and asexual populations of Chlamydomonas reinhardtii growing on various combinations of carbon sources (Colgrave et al., 2002). More recently, Goddard et al. (2005) constructed an asexual yeast strain by deleting the genes SPO11 and SPO13 that are required for meiosis and recombination. This asexual strain was able to experience a cycle of experimental environments, including a sporulation medium, identical to that of the isogenic sexual strain, but did not go through meiosis and was only able to produce diploid spores. When these strains evolved independently in a harsh environment (glucose limited, temperature and osmolarity elevated), the sexual population adapted more quickly than their asexual counterparts, whereas neither subpopulation changed fitness in a benign environment to which they were already well-adapted. The experiments presented here use similar strategies within the yeast experimental system to test several theories on the benefits of sexual reproduction: 1. The importance of spatial structure to the benefits of sexual reproduction. Spatial structure adds a biologically realistic component to the Fisher-Muller hypothesis, which I hypothesize can make recombination more likely to break down adaptive allele combinations than to construct them. 8

15 2. The importance of segregation and homozygosity of beneficial mutations, rather than recombination, in harsh vs. benign environments. 3. Experimental tests of the principles and predictions of another hypothesis for the advantage of sex, the Tangled Bank. This hypothesis is based on the premise that more genetically diverse offspring are more ecologically diverse, and therefore able to exploit available niches more productively. It was proposed in the 1980s, and has been largely ignored since then, but has recently received renewed theoretical interest, but almost no empirical testing. 9

16 CHAPTER 2: SPATIAL STRUCTURE AND THE EVOLUTION OF SEX INTRODUCTION The genetics-based process of periodic selection, in which increasingly fit genotypes replace preceding ones, and the ecological process of competitive exclusion on a single limiting resource (Hardin, 1960), are both expected to remove genetic variation within strictly asexual populations. In simple, novel environments where adaptation is expected to occur, asexual populations are traditionally expected to be at a disadvantage because adaptive mutations arising at the same time in different lines could not be combined as in sexual populations. Instead, adaptive mutations at different loci would compete with each other until only the fittest remained, and combinations of beneficial mutations could only be obtained if they occurred in succession within the same lines of descent (Muller, 1932). In contrast, new mutations could be combined in sexual populations and hasten the fixation of both. Although this still results in a decrease in genetic diversity, sex would speed adaptation by fixing new mutations quickly. Experimental support for sex hastening adaptation to a novel environment has been documented by comparing cell division rates (fitness) in separate sexual and asexual populations of the unicellular alga Chlamydomonas reinhardtii growing on various combinations of carbon sources (Colgrave et al., 2002). Additionally, Goddard et al. (2005) found that after 300 vegetative generations (punctuated with sporulation) in a chemostat, sexual populations of yeast adapted faster than otherwise isogenic asexual populations to a harsh (high osmolarity and temperature) environment. The expected inability of asexual populations to maintain genetic diversity is a central focus of evolution of sex theories, but there is evidence that selection can 10

17 maintain, rather than eliminate variation. There are an increasing number of studies, both theoretical and experimental (Turner et al., 1996; Rozen and Lenski, 2000; Rainey and Travisano, 1998; Rainey et al., 2000; Campbell, unpublished thesis) that show that ecological diversity can evolve and persist within asexual populations. These studies provide evidence that is in direct contrast to competitive exclusion and the selective sweep model of adaptation, which suggest that selection is always for a single competitively superior genotype that is replaced in the next sweep by a better competitor. This implies that there is no ecological variety, but ecologists have known for some time that natural communities do not evolve in this way. In the real world, adaptations usually come with tradeoffs, and diversity is maintained through processes such as frequency-dependent selection and through spatial and temporal variation in the environment. With ecological variation it is unlikely that there is a single competitively superior genotype in each population, since fitness would depend on factors such as resource patchiness, interactions with other genotypes, spatial structure, etc., and diversity would persist. The following are examples of such diversification in microbial evolution experiments. Rozen and Lenski (2000) detected a polymorphism of large and small colonies that arose in an Escherichia coli population founded by a single clone and propagated for almost 20,000 generations in a glucose-limited culture. They found that the polymorphism was likely maintained through frequency-dependent mechanisms such as cross-feeding, in which one genotype can metabolize the secretions of another, and through differing death rates at separate phases of the daily growth curve. Friesen et al. (2004) propagated E. coli in batch culture in a resource mixture of glucose and acetate 11

18 and found that five out of twelve starting populations became clearly bimodal for colony size. The availability of two carbon sources allowed diauxic growth in which a sugar is fermented first, followed by respiration of carbon sources, such as acetate, that cannot be fermented. Colony morphology often corresponded with differing patterns of diauxic growth (the metabolic shift from fermenting sugars to respiration), providing an explanation for the stable polymorphism. The inclusion of both glucose and acetate from the beginning of the experiment made ecological divergence more probable than in other experiments where only one carbon source was provided. In those previous experiments, cross-feeding could only arise from byproducts of the metabolism of that single carbon source. Another example of diversification in asexual populations comes from Habets et al. (2006) who propagated replicate lines of E. coli, started from a single strain, in one unstructured population (in liquid culture) and two structured populations. Adaptive radiation, catabolic diversity, and frequency-dependent fitness interactions were only evident in conditions with spatial structure (grown on agar plates) and intact population structure (no dispersal). Rainey and Travisano (1998) found that the aerobic bacterium Pseudomonas fluorescens diversified into distinct morphological classes when propagated in spatially structured (unshaken) cultures, but did not diverge when the spatial structure was disrupted by shaking. These experiments show that ecological variation is possible even in tubes and on plates in the lab, allowing us to draw conclusions about diversity in the natural world. The previous examples are especially interesting because ecological divergence was observed in asexual populations started from a single clone. The fact that neither 12

19 lineage in those experiments acquired a mutation that enabled it to drive the other extinct (over ~18,000 generations in one case (Rozen and Lenski, 2000)), implies that their ecological specializations are mutually exclusive. When an environment provides these opportunities for specialization, then it might favor an asexual population rather than a sexual one due to strong synergistic epistasis, a condition in which recombinants are less fit in either niche than individuals specialized for one of them. All of the aforementioned experimental populations were purely asexual, and I aimed to provide the first, to our knowledge, test incorporating sexual populations. Sex is hypothesized to be beneficial because it facilitates the spread of advantageous alleles by allowing them to escape from their initial genetic backgrounds (Fisher, 1930) and because recombination can combine adaptive mutations from separate lineages, allowing them to be selected for simultaneously rather than competing (Bell, 1982). Although this has been demonstrated experimentally when the emphasis is on the abiotic components of the selective environment, when interactions with nearby genotypes are an important component of the selective environment, co-evolution and specialization may be adaptive. In this case, sex may impede adaptation by producing offspring with combinations of alleles that are not well-suited to either of the roles for which their parents have specialized. Thus, population structure, or the lack thereof, may determine if and when sexual reproduction is advantageous, and intact population structure is expected to favor asexuals. Here I describe an evolution experiment in which otherwise identical sexual and asexual yeast compete and adapt on agar plates of glucose-limited minimal medium, a new selective environment. Unlike liquid cultures, agar plates allow the manipulation of 13

20 population structure, specifically prevention or enforcement of dispersal. Prevention of dispersal is predicted to favor ecological specialization, and therefore an increase in the frequency of asexuals, as recombination among sexuals would impede specialization by breaking up synergistic allele combinations. On the other hand, if potential localized ecological interactions are repeatedly disrupted by dispersal, selection would favor general competitive fitness against a variety of ecotypes, and the ability to recombine mutations that are generically adaptive on glucose-limited minimal agar may be an advantage for the sexual subpopulation, although previous experiments have found no such advantage in liquid media of similar composition (Goddard, 2005; Baliga, unpublished thesis). In order to induce meiosis, yeast populations must be nitrogen-starved with nonfermentable acetate as the carbon source. Experiments evolving sexual and asexual populations separately require repeatedly imposing on the sexuals this potentially confounding selective factor that is not applied to asexual populations. An isogenic knock-out created by the replacement of the IME1 gene, which regulates and is necessary for meiosis (Mitchell, 1994; Kassir et al., 1988), with an antibiotic marker in the asexual subpopulation enabled the sexuals and asexuals within each treatment of this experiment to adapt under identical conditions. Within each group of dispersed or not dispersed plates, controls were maintained without sporulation, which is the meiotic stage in yeast and is the time when novel genetic combinations would be produced. No consistent changes in frequency were expected for the potentially sexual subpopulations in the populations that were not allowed to undergo sporulation 14

21 METHODS Construction of asexual and sexual strains Primers with nucleotides of sequence identity to the ends of the IME1 target gene (Table I) were used to amplify the KanMx4 and Pag32 cassettes, which confer resistance to the antibiotics G418 and hygromycin, respectively (Wach, et al., 1994). Saccharomyces cerevisiae strains DH45 and DH46, which are derived from Y55 (McCusker and Haber, 1998) were transformed through a lithium acetate protocol (Golemis, 1994) and a diploid homozygous for the knockout was constructed by mating two haploid strains: ime1::kanmx4 and ime1::pag32. Asexuality of the knockouts was confirmed when no asci were seen after growth on dextrose-limited minimal medium (DM) agar plates (0.17% yeast nitrogen base without amino acids, 0.5% ammonium sulfate, 60 mg/liter leucine, 0.25% glucose, 2% agar) and transfer to zinc acetate sporulation plates (1% potassium acetate, 60mg/liter leucine, 50mg/liter zinc acetate, 2% agar). The neutrality of the marker was tested through a standard 48 hour competition as described by Zeyl and DeVisser (2001). The diploid transformant chosen for use in the long-term competition had a relative fitness of ± (mean +/- 95% confidence interval) when competed against the sexual ancestor. Outcrossing efficiency of sexual ancestor The effect of sexual reproduction being examined-combining mutations from different individuals--nly exists if mating occurs between spores from different asci, rather than among spores within an ascus, which are derived from the same diploid cell by meiosis. To confirm that outcrossing was taking place, the outcrossing efficiency of 15

22 the sexual ancestor was determined by separately mixing the sexual ancestor (leu2δ, URA3) with either a MATa, LEU2, ura3δ or a MATα, LEU2, ura3δ strain and then allowing sporulation of, and mating within, the mixed cultures. The fraction of cells that were outcrossed was calculated by dividing the number of colonies on plates of unsupplemented minimal medium by the number of colonies on plates containing both uracil and leucine. The average fraction from five replicates of outcrossed cells over ten days was calculated for each mating type and the averages were summed to give an estimated outcrossing rate of 50.44%. Experimental conditions Sexual and asexual subpopulations were mixed at equal proportions and plated without dilution onto DM plates with 10 replicate populations for each of the four treatments. Selection occurred when the populations were transferred onto new media every 48 hours by replica-plating. Half of the populations were induced to sporulate once every two weeks and the frequency of sexuals was assayed each week. The populations were allowed to evolve under four different conditions: 1. Structured, meiosis not induced: colonies transferred to the same relative positions on new plates and never placed on sporulation plates. 2. Structured, meiosis induced: colonies transferred to the same relative positions on new plates and placed on sporulation plates once every two weeks. 3. Dispersed, meiosis induced: repeated transfers from the old plate onto a new plate with the old plate rotated counter-clockwise at a random angle between transfers (following the protocol of Kerr et al., 2002) to disperse the colonies onto the new plate in new spatial arrangements with respect to other colonies. 16

23 There were two of these rotations during each transfer and these colonies were never placed onto sporulation plates. 4. Dispersed, meiosis induced: rotation occurred as above, but the colonies were placed onto sporulation plates once every two weeks. Consistent pressure during replica-plating was achieved by using a second replica-plating tool as a weight (Figure 1). Sporulation was induced in the two sets of non-control populations once every two weeks by replica-plating onto sporulation plates. The experimental schedule was as follows: Day 1: Replica-plate from sporulation plates (experimental populations) or DM plates (control populations) onto fresh DM plates. Days 3, 5, 7, and 9: Replica-plate (with or without dispersal) onto fresh DM plates. Days 11-14: Transfer each experimental population to sporulation agar and room temperature for 4 days. Since control populations remained on their DM plates during the sporulation period, the sexual subpopulations were not induced to sporulate, and no additional growth occurred during the additional period they spent on DM while other sexual subpopulations underwent sporulation. The frequencies of sexuals were assayed on Day 3 and Day 9 (either before or after sporulation) by adding 1 ml of sterile water to the plates and scraping the cells into a 10 ml water tube. The culture was then diluted and plated onto permissive DM plates followed by replica-plating onto selective G418 plates. 17

24 Generation time The number of generations produced during each 48-hour period was estimated by mixing equal volumes of the sexual and asexual subpopulations grown overnight at 30 C in DM liquid media and plating without dilution onto five DM plates. On the third and fifth days, each plate was replica plated onto a new plate. On the seventh day, each plate was replica-plated twice and the cells from one set of plates were washed off and onto new dextrose-minimal plates, which were then counted after two days of incubation to get an initial cell count. The remaining plates were incubated for another 48 hours after which the cells were washed off, diluted, and plated onto new plates to get final cell count. After adjusting for the dilution factor, the number of generations for each pair of plates was calculated as log(2)(final pop size/initial pop size). The average number of generations per 48 hours was estimated to be 11.56, so the evolution experiment produced approximately 47 mitotic generations and one meiotic generation within each two-week cycle. Estimates of competitive fitness of evolved isolates in experimental environment Two sexual and two asexual (resistant to the antibiotic G418) colonies were selected at random at the end of the experiment from each of two populations for each of the four treatments After transfer to liquid culture, the 32 isolates were frozen in 15% glycerol at -80 C. Competitive fitness was again determined as described by Zeyl and DeVisser (2001), but with the acclimatization and competition occurring on DM plates instead of in liquid DM. 18

25 Estimates of sporulation frequencies The ending plate populations were scraped into liquid culture and frozen in 15% glycerol at -80 C. Frozen culture aliquots of 100μl each was plated onto a minimal medium plate. After one cycle of replica-plating, with or without rotation, the control populations were left at room temperature and the populations with meiosis induced were replica-plated onto sporulation plates. After 4 days, the number of sporulation events out of 100 cells was counted for each population. Statistical analysis The frequency and fitness data violated the requirements for parametric tests. Once log transformed, ANOVA was used to determine dispersal and sporulation effects and their interactions on ending frequencies and fitnesses. 19

26 Table I: Forward and reverse primer sequences for replacing the IME1 gene. Forward primer Reverse primer TAATAAAAGAAAAGCTTTTCTATTCCTCTCCCCACAAACAAAATGCAGCTGAAGCTTCGTACGC ATATTTTGAGGGAAGGGGGAAGATTGTAGTACTTTTCGAGAATTAGCATAGGCCACTAGTGGATCTG 20

27 No dispersal Dispersal Figure 1: Replica-plating scheme. Top depicts single direct transfer (without rotation) used for populations without dispersal. Bottom depicts transfer with rotation causing dispersal of colonies. 21

28 RESULTS After approximately 1000 generations (~298 days), and 16 sexual cycles in populations in which meiosis had been induced, trends in the frequencies of sexuals were as predicted in three of the four treatments (Figure 2). The frequencies of sexuals decreased in the spatially fixed (no dispersal) populations in which meiosis was induced, but remained close to 50% in the corresponding control populations. Surprisingly, the frequency of sexuals increased in both the meiotic and ameiotic (control) treatments of the dispersed populations. Both dispersal and sporulation had significant effects on the ending sexual frequencies (Table II). This supports the hypothesis that the potential benefits of sexual reproduction can be influenced by spatial structure. Given the difference in frequencies of sexuals and asexuals between the dispersed and non-dispersed experimental populations, and our expectation of specialization in the latter, we set out to determine if there were also noticeable fitness differences between the treatments. Assays of isolates at the end of the experiment showed that fitness, as measured by competition against the ancestor, increased in the dispersed populations, but not in the non-dispersed ones (Figure 3). There was a significant difference in average fitness between the isolates from the dispersing populations and those from spatiallyfixed populations (Table III). The increase in frequency of sexuals in the control (without meiosis induced) populations that had undergone dispersal was unexpected because those populations never experienced the environment that induces sporulation in yeast. One explanation is that the sexuals were undergoing meiosis even when it was not induced. To test this explanation, we assayed the number of asci present in samples from the ending 22

29 populations. Control populations were assayed on DM, which is not generally known to induce meiosis in S. cerevisiae, and this assay confirmed the occurrence of unexpected sporulation (Figure 4) in both the non-dispersed and dispersed control populations. 23

30 Frequency of sexuals 1.0 frequency of sexuals no dispersal dispersal Figure 2: Frequencies of sexuals in the ending populations evolved on agar plates with and without rotation. Symbols are replicate populations and bars show means of replicate populations for each treatment. Gray circles are controls, in which meiosis was never induced. Black circles represent populations in which meiosis and sporulation were induced. 24

31 isolate a s a s a s a s no dispersal dispersal Figure 3: Fitness of random isolates from evolved populations, estimated by competition against the ancestral strain. For each dispersal treatment, two sexual and two asexual isolates were sampled from each of two control populations (gray symbols) and two experimental populations (meiosis induced, black fills). Each column of symbols shows five replicate fitness estimates for each of two sexual isolates, labeled s on the X axis (circles for one isolate, triangles for the other, and a bar for their pooled mean) or two asexual isolates, labeled a, from one population. The fitness of the ancestor is 1.0 by definition. 25

32 40 30 numer of asci no dispersal dispersal ancestors Figure 4: Numbers of asci out of 100 cells in each population after one experimental cycle. Gray symbols represent isolates from control populations assayed after incubation on glucose-limited medium. Black symbols represent sexual isolates from populations in which meiosis was induced 16 times during the evolution experiment, incubated for this assay on sporulation medium to induce sporulation. 26

33 Table II: Two factor ANOVA of the effects of dispersal, sporulation, and their interaction on the final frequencies of sexuals in the ending populations. Sum of Source squares df Mean Square F Prob > F Dispersal P < Sporulation P = Dispersal * Sporulation P = Error R 2 = 0.638, adjusted R 2 =

34 Table III: Three factor ANOVA of the effects of dispersal, sporulation, sexual ability (whether an isolate was IME1 and capable of meiosis, or incapable of meiosis due to IME1 deletion) and their interactions on the final fitness of isolates from populations evolved on agar plates. Source Source SS Mean Square df F Prob > F P = Dispersal Sporulation P = 0.80 Sexual P = 0.31 Sporulation*Dispersal P = 0.79 Sexual*Dispersal P = 0.75 Sexual*Sporulation P = 0.64 R 2 = , adjusted R 2 =

35 DISCUSSION My results provide evidence that fixed population spatial structure can select against sexual reproduction even in an environment where dispersal favors sex. The reduced success of the sexuals on the plates without dispersal supports the hypothesis that the opportunity for the evolution of interactions other than competition on those plates favors divergent genotypes rather than the hybrids that would be produced by sexual reproduction. In contrast to the dispersed populations, fitness did not increase after 1000 generations in the non-dispersed populations, a striking and otherwise hard to explain result that is consistent with the hypothesis that with ecological interactions restricted to persistent local associations, selection would favor the genotypes that benefitted most from those localized interactions rather than the best competitor on DM agar. Sexual reproduction appeared to be favored when localized interactions between cell lineages were frequently disrupted by dispersal. Although the dispersed and non-dispersed populations that underwent meiosis behaved as expected, my overall conclusion is weakened by the unexpected behavior of a control treatment that did not have meiosis induced. Without sexual cycles, the frequencies in all of the control treatments were expected to either remain at approximately 50% sexuals and asexuals, or for one subpopulation or the other to approach fixation in most replicate populations due to hitchhiking on adaptive mutations. However, the frequency of sexuals increased on almost all of the dispersed plates without meiosis, which fits neither expectation. This can be at least partially explained by the presence of asci (the products of meiosis and sporulation) in some control populations on a medium where none were expected (Figure 4). It has been shown that infrequent sex can be equally effective as obligate sex at reducing linkage equilibrium (Green and 29

36 Noakes, 1995), and it is possible that this low level of sexual recombination, combined with dispersal, was enough to affect the frequencies of sexuals in the control population. Although asci were also seen in the non-dispersed control populations, sexual reproduction was not advantageous in those conditions, so the frequency of sexuals did not increase. Overall, the data suggest that sex is favored in an environment in which biotic interactions are always changing. The average fitness of isolates from the dispersed populations was higher than that of the isolates from the non-rotated plates. This is consistent with the hypothesis that localized interactions in the non-dispersed populations contributed to adaptation and fitness in a way that we were unable to preserve when selecting isolates for competitions. It is also consistent with Kryazhimkiy et al. s (2012) finding that adaptation slows in isolated populations of S. cerevisiae when compared to well-mixed populations. They propagated asexual yeast populations in YPD with daily serial transfer, at various levels of subdivision/migration. After 550 generations they found that the well-mixed populations had the highest fitness increase, and the isolated demes had the lowest. This was expected in my experiment if the colonies on the rotated plates had evolved some sort of cross-feeding or other interaction. 'The competitive equality of isolates from structured populations to their ancestor despite ~1000 generations of selection is an otherwise remarkable example of stasis that supports our expectation of ecological tradeoffs between competitive fitness and proficiency in exploiting a coevolving biotic setting. We theorize that some division of resources or cross-feeding is involved as in Rozen and Lenski (2000). Dispersal created more ecological variation and variation in selective pressure, which led to a general 30

37 increase in fitness on DM, rather than an increase in fitness that was dependent on coevolution. This supports the theory that sex is beneficial in changing environments, whether the changes are biotic or abiotic. It is also possible that the mutations that were selected for in the structured environments result in lower fitness when combined in the same individual. There is experimental evidence of this reciprocal sign epistasis in evolving yeast populations (Kvitek and Sherlock, 2011), in which certain mutations are mutually exclusive because they are ecologically incompatible. They used five of the fittest clones from a yeast population evolved in 0.08% glucose for 448 generations in a chemostat culture. Through sequencing, backcrosses, and competitions, they were able to determine that mutations in MTH1 (a negative regulator of the glucose-sensing signal transduction pathway) and HXT6/HXT7 (a high-affinity glucose transporter) were individually adaptive in the glucose-limited environment, but the fitness cost of the double mutant prevented them from being selected together, and resulted in a fitness valley. In our case, these combinations of incompatible mutations could be produced through recombination rather than successive mutation. Although my experiment was conducted on small scale, the relationship between sexual reproduction and changing environments can have greater implications when considering habitat fragmentation and geographic isolation of populations. For example, Noth et al. (2010) used a spatially explicit, stochastic model that included parameters such as patch creation, local adaptation, dispersal, and mutation, and found that sexual reproduction is one factor that can act against local adaptation in a stationary landscape. If the founding population is diverse enough, sexual reproduction can initially accelerate 31

38 local adaption, but will inhibit adaptation in the long run, unless the landscape changes. This model suggests that however prevalent sex may be, there are conditions in which it is less than favorable. My results confirm that spatial structure, and the constant environment it provides, is one of these cases. 32

39 CHAPTER 3: MATING TYPE SWITCHING, SEGREGATION, AND THE ADVANTAGES OF SEXUAL REPRODUCTION IN HARSH ENVIRONMENTS INTRODUCTION The advantages of recombination have long been considered to be vital to the explanation of the evolution and maintenance of sex. Although recombination is most often cited as the main advantage of sexual reproduction, the advantage of segregation may be all that is required. Kirkpatrick and Jenkins (1989) noted that in an asexual diploid population, two separate mutations at the same locus are required to achieve homozygosity, whereas in a sexual population, segregation can carry a single mutation to the homozygous state. Otto (2003) has shown that in model populations with a low level of inbreeding, selection is stronger on a modifier allele that promotes segregation than on a modifier allele that promotes recombination. The model tracked the frequencies of alleles at two loci: a modifier allele that determined allocation to sexual reproduction vs. asexual reproduction, and a fitness allele that could be under various forms of selection. With random mating, sexual reproduction was only favored under selective conditions that also favored recombination and selection on the modifier was very weak. With inbreeding, however, selection on the modifier of sex was stronger even when selection on the fitness loci was weak, as long as deleterious alleles were partly recessive. Therefore, it may be the case that homozygosity, rather than (or in addition to) interactions between loci is responsible for the maintenance of sexual reproduction. There have been numerous experiments addressing the potential benefits of sexual reproduction, but usually with only recombination in mind. Colegrave (2002) determined 33